Skip to Content

Business and Markets

One Of Those Binary Events for AbbVie

Back in late 2015, I wrote about cancer stem cells, the idea that some tumors are driven by populations of slow-dividing stem cells. The theory has been that these are always going to be in there fighting against your efforts to kill off the downstream tumor cells, so unless you find a way to target them, you’re not going to be successful. As that blog post shows, though, this area was imperfectly understood at the time, and it has not gotten a great deal more clear since then.

The company that I wrote about, though, Stemcentrx, was bought a few months later by AbbVie for 5.8 billion dollars in cash (and promises of about four billion more if things worked out). At the time, I wrote (uncontroversially, I think) that “this deal is going to end up looking either very smart or very stupid”. The lead program out of this deal, Rova-T, is an antibody-drug conjugate (rovalpituzumab tesirine) which was aimed at small-cell lung cancer (SCLC), specifically the DLL3 protein found on the surface of some cell types. The antibody targets DLL, and the conjugate has a cleavable linker that delivers a pyrrolobenzodiazepine dimer as a DNA-damaging payload. SCLC is bad news, so any advances in this field would be welcome.

Later on in 2016, AbbVie presented data from a small trial (I believe this one) that showed that response rates did seem to correlate with DLL3 expression. But the rates themselves were not all that impressive, in the larger scheme of things, and overall, people didn’t seem very impressed. AbbVie pushed on, though, seeking accelerated approval from the FDA, and today we have the latest results. Unfortunately. The company reported the response rate from the DLL3-high patient group, and it’s only 16% (which is the same or worse than standard of care response rates), and the survival data are not impressive, either. (Update: it’s quite likely that this is even worse than it seems!) The stock market is not taking this well. ABBV is down $12 a share as I write, which takes out nearly $20 billion dollars of market cap from the stock. The fear is that if this is the DLL3-high group, which is supposed to be the best of the bunch, what are the full Phase 3 results going to look like? Forget about accelerated approval – AbbVie sure has to – what are the chances that this program works well enough to be approved at all?

This also puts a bit of a hole in the cancer-stem-cell idea in general. It’s a complicated enough field that you’re not going to disprove it with a single agent or in a single type of cancer, but this does show that it’s not even as straightforward as it looks (and it didn’t look that straightforward!) As for this specific approach, and this specific deal, well. . .I suppose things could change eventually, but for now, it’s a disaster. We’ll see what AbbVie does after the full data come out. . .

Update: Matthew Herper’s take on this is well worth reading. I didn’t do the sack dance on Peter Thiel in this post, but it’s not completely inappropriate. . .

18 comments on “One Of Those Binary Events for AbbVie”

  1. Anon says:

    As always too much hyped information lead to false hope except that it also bites into earning. Cancer area has simply too much to offer in terms of treatment options, some novel and other over hyped one. Buyer beware!

  2. I think you mean ‘slow-dividing stem cells’ — the theory being that conventional chemotherapy kills the bulk tumor (because they are fast dividing) but spares the more important stem cells.

    1. Derek Lowe says:

      That I did! Fixed. . .

  3. Tom Price says:

    Peter Thiel

    1. Isidore says:

      Peter Thiel made his money when Abbvie bought Stemcentrx, no?

      1. Derek Lowe says:

        Most definitely – he and the Founders Fund itself did very well.

      2. tangent says:

        Gee, I wonder why they didn’t pull for some compensation in AbbVie stock in lieu of the cash. Or just a straight cut of the drug’s eventual sales.

  4. Hap says:

    Pharma has been buying back lots of stock because they can’t see how to invest in developing their own drugs and make money doing it (though, to be logical, that suggests that either the people managing shouldn’t be there because they don’t know what they’re doing or the company shouldn’t be spending any money because it’s impossible to make money in their business and so any spent money is just decreasing the assets that investors could use to make money doing something else). If you don’t know what to invest in with full internal knowledge of your research and its possible financial implications, what makes you think you can find drugs elsewhere with limited data (companies don’t have an incentive to get data that could lower their ability to sell out)?

    It seems like pharma is playing a sucker game with other people’s money. If you can’t make money running your business, then perhaps you shouldn’t be running it. If you can’t afford to wait for your money, don’t invest in long-term businesses; if you can’t wait for the buns to come out of the oven before you eat or sell them, then maybe you shouldn’t be a baker.

    1. PV=nRT says:

      It doesn’t scale up very well. Adding a bunch of money to RnD budgets with existing teams just adds more administration problems and may not add good data or compounds. If you want to have an impact, you are probably better off buying the results of a small team’s work, or funding a new collaboration.

      1. Hap says:

        But isn’t that the thing? If you don’t know what to work in or can’t work well (so that you can’t actually get results), then either you are doing something wrong (because you can’t pick good things to work on or recognize them when they come up) or you’re doing something that no one else can do, either. If you can’t recognize useful things with complete (well, as complete as you can get) information, why do you think you can do so with less complete information (and incentives to make it less complete)?

        A separate issue is that that model seems to limit the pool of useful data – use once and discard means that few people (those who can tolerate the risks or have enough money to mitigate them) are going to stay around to get pools of data from which to get better information (of course, the counter is that the data isn’t helping anyway, but in general getting less data and wisdom never works well). If you’re the NFL, you can get people to spend their lives for a chance at fame and a big payoff. If the payoff’s not there (and for almost everyone, even if their companies win, it won’t be), then neither will the people to do the work.

        You can pay for organization (someone else’s) but if you’re going to do something and you know you can’t do it well (that your organization prevents you from doing it well), fixing it would seem to be a good idea.

        1. PV=nRT says:

          So the question would be, What’s the best way to fund pharma innovation?, then.

          Having been at a big pharma that was spending literally 50% of it’s budget on two pet projects that were never going to be drugs, I won’t advocate that that model is perfect.

          On the other hand, there are certainly Sirtris type mess ups when big pharma goes out and is determined to make deals when there’s nothing good to be bought. This isn’t always due to asymmetrical information, or to lack of aggressive information-seeking on the part of the seller. It’s often due to just not knowing whether there could ever possibly be clinical success until you get there. This is often not answerable except in the clinic, unfortunately. Except for me-toos, that’s almost all we do.

          Buying back stock is pointless and should be illegal, as should many wall St practices.

  5. steve says:

    The stem cell hypothesis was an attractive one at the beginning – namely that there was a subpopulation of cells that self-renew and give rise to the bulk population of cancer cells towards which most conventional chemotherapies were directed. This was proven to be true for leukemias (particularly by John Dick) using well-defined hematopoietic stem cell assays. The concept began to fall apart with solid tumors. A standard “proof” was to do limiting dilution analyses and see what percentage of tumor cells would give rise to tumors in immunocompromised mice. However, Sean Morrison showed that this was highly dependent on how immunocompromised the mice were – in nude mice it looked like there was a stem cell population but in Rag3 KO mice, with very little in the way of an immune system, virtually 100% of the cells in melanoma were tumorogenic. This gave way to a bunch of ad hoc definitions (always the death knell of a theory) saying that the stem cell population in a tumor could vary from <1% to 100% of the cells. Of course, this always was nonsense since it up ends the whole meaning of what a stem cell population is.

    What probably happens is that tumor cells are plastic and can take on stem cell phenotypes as selective pressures increase. Rather than a discrete stem cell population we should be applying principles learned from stem cells (self-renewal pathways, multi-drug resistant pathways, quiescence, niches, etc) to all cancer cells. The original idea that you need to just knock out a discrete stem cell population is likely not true for solid tumors but the idea that you need to target stem cell pathways as well as those present in bulk tumor cells probably is.

  6. Chrispy says:

    Someday the antibody-drug folks are going to have to explain why the drugs are simply not working very well. Period. We have Adcetris, Mylotarg (back again!), Besponsa and Kadcyla after decades of effort and many, many failures. The concept is so attractive that we keep dumping energy into it, even as the evidence piles up that it is not a robust, generalizable approach.

    1. Imaging guy says:

      I agree. Now we are not even sure whether antibody drug conjugates (ADC) need to enter the cells to exert their effects. Some are saying the payload can be cleaved outside the cells.

      “Antibody drug conjugates and bystander killing: is antigen-dependent internalization required” Br J Cancer. 2017 Dec 5;117(12):1736-1742. PMID: 29065110

    2. yf says:

      The trickiest part of ADC is the linker. A lot of enzyme based linkers are tested in vitro with either purified proteins or cell lines. Efficacies obtained from these systems may be far from the realities in vivo. I personally do not believe any enzyme will be constitutively active in vivo. Myriads of inhibitors and other regulatory system will prevent that from happening (e.g. complement system, MMP inhibitors). Failure to consider the inhibitory system, depending solely on positive analytical data obtained from biochemical tests using limited numbers of purified components, and blissfully ignore the complicated design of a biological system warrant disappointments.

    3. Mr T says:

      ADCs already save lives in ways that no other drugs can. Just for that, it was worth taking the time to do the research. The rules of ADCs are just beginning to be understood and formulated. The number of factors involved is much higher than in classical small molecules discovery. Once these rules are refined, I predict bright days ahead for ADCs. I agree with you, however, that the concept will have to target specific patient populations in order to obtain the best outcomes.

  7. Tips for PIs says:

    Learn to be humble. Just b/c you are a PI doesnt mean you know very much. Grad students are not your kids, not even your academic ones: most of them are working on their own problems that will eclipse your ideas big time. You have far more to learn from them that they do you, unlike real parents/kids. You have a lot of growing up to do. Thx.

    1. Bendohellar says:

      Censored it again. God, for someone that talks out his own hole so much you sure do love to slet other people up dont you buddy?? Truth does not exist, as you have proven wonderfully. It all over.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.